JOHN WILLIAMS’ SHADOW GOVERNMENT STATISTICS

COMMENTARY NUMBER 265
November Retail Sales, Inflation Surge, Data Distortions

December 13, 2009

__________

Renewed Caution on Depression-Warped Data

November Retail Sales Annual Gain of 1.9%
Reflected Return of Inflation

November Annual CPI Inflation
Should Jump by About 2%

 __________

PLEASE NOTE: Traveling on Friday, I found that the full data needed for this Commentary were not available on a timely basis, hence the weekend posting. I apologize for any inconvenience. The next scheduled Commentary is planned for Friday (December 16th), following release of the November CPI and housing starts, including the prior day’s release of the November PPI and index of industrial production. The preliminary SGS Ongoing M3 estimate for November and the November U.S. dollar indices have been updated on the Alternate Data tab at www.shadowstats.com.

– Best wishes to all for a most joyous holiday season!  John Williams


Depression Distortions of Economic Reporting Intensify. A continued note of caution is offered on all current economic reporting as data distortions generated by the unusual length and depth of the ongoing economic downturn. The economic crisis has been of a severity that is unprecedented in the post-World War II era of modern economic reporting. None of the government’s statistical reporting agencies have been through this before.

Beyond special factors in various series, such as an extreme level of foreclosure activity impacting housing data, one-time stimulus gimmicks (i.e., cash-for-clunkers boosting auto sales), or the GM and Chrysler bankruptcies throwing off regular automobile production cycles, basic models are crumbling and methodologies such as seasonal-factor adjustments are generating distorted headline numbers.

Adjustments for seasonal factors are designed to remove the impact of regular patterns of activity that tend to repeat each year. The seasonal factors usually are most heavily weighted by the prior year’s activities, but what was happening in the big swings tied to the economic free-fall last year were not seasonal in nature. The effect on current economic data, though, will tend to reflect a weaker seasonal patterns one year ago, which will boost artificially the adjusted data being reported at present. The Fed and the Bureau of Labor Statistics (BLS) have acknowledged difficulties in these areas. Separately, the BLS effectively has acknowledged that its birth-death model — used in setting the reported levels of and month-to-month changes in payroll employment — is broken.

The retail sales and trade series are among those subject to these general issues, but the distortions often are not obvious.

Series Revisions and Surging Inflation Clouded November Retail Sales Reporting. Boosted by higher monthly gasoline prices, and an annual surge in same, Friday’s (December 11th) November retail sales report — issued by the Census Bureau — indicated a statistically-significant, seasonally-adjusted  monthly increase of 1.29% (1.32% net of revisions) +/- 0.6% (95% confidence interval). Such followed a downwardly revised 1.14% (previously 1.37%) monthly gain in October. 

On a year-to-year basis, the November year-ago comparison was against collapsing gasoline prices and gasoline station sales, the opposite of current trends.  Accordingly, November 2009 retail sales were reported up by 1.90% from November 2008, but that was thanks also to a boost from an unusually large downward revision to the November 2008 numbers.  Against the November 2008 retail sales in place with last month’s report, the November annual gain was 1.26%. The November annual gain followed a downwardly-revised annual decline of 2.01% (was 1.74%) in October.   In any event, with the impact of inflation removed, even the spiked November annual gain should be flat to minimally minus.

New Sampling Leaves Monthly and Annual Data Somewhat Non-Comparable. Every 30 months, or so, the Census Bureau revamps its sampling universe for the "advance" retail sales report, which is what provides the headline number on the series. Without a chance of seeing how earlier reporting would have fared with the new sampling, there is no way of telling to what extent the new universe may be impacting or biasing the headline monthly and annual changes.

Real Retail Sales. Removing the effects of inflation, November retail sales activity should show a monthly gain, but an annual change that is flat or in contraction. The pattern of ongoing, inflation-adjusted activity remains one of bottom-bouncing/plateauing at extremely low levels. Details will be updated and graphed with the Commentary following the November CPI release on Wednesday (December 16th).

Core Retail Sales.  A change in "core retail sales" methodology was introduced three months ago, where the net relative monthly increases and/or decreases in gasoline station and grocery store sales were subtracted from the full monthly retail sales number, instead of the total of gasoline station and grocery store sales each month. Assuming that the bulk of non-seasonal variability in food and gasoline sales is in pricing, instead of demand, the revamped reported "core" change more closely reflects the actual retail sales experience. This remains a work in progress and eventually will be used in the development of additional SGS alternative economic measures.

For the near-term, the "core" retail sales is reported in two versions, where Version I uses the original methodology, and Version II version appears to provide a more balanced picture of the impact food and energy inflation in the standard retail sales reporting.

Consistent with the Federal Reserve’s predilection for ignoring food and energy prices when "core" inflation is lower than full inflation, "core" retail sales:

Version I — November retail sales net of total grocery store and gasoline station revenues — rose by 0.8% (0.9% net of revisions) versus the official aggregate gain of 1.3%.    

Version II — November retail sales net of the monthly change in grocery store and gasoline station revenues — rose by 0.6% (0.7% net of revisions) versus the official aggregate gain of 1.3%.

Reported October Trade Deficit Narrowed. The Census Bureau and Bureau of Economic Analysis reported that the seasonally-adjusted October trade deficit narrowed to $32.9 billion from a revised $35.7 (previously $36.5) billion in September. The trade improvement reflected a fall in the average price of imported oil from $68.17 per barrel in September to $67.39 in October, while physical imports showed an unseasonably sharp decline from 9.5 million barrels per day in September to 8.3 million barrels per day in October. In 2008 the daily barrels-per-day import rate soared from 8.4 million in September to 10.3 million in October, a pattern that certainly cannot be easy to stabilize in terms of regular seasonal activity. On top of seasonal factor distortions, it still appears as though irregular paperwork flows through Customs are impairing the reporting accuracy of imports.

Week Ahead. Given the underlying reality of a weaker economy and a more serious inflation problem than generally is expected by the financial markets, risks to reporting will tend towards higher-than-expected inflation and weaker-than-expected economic reporting in the month ahead. Such is true especially for economic reporting net of prior-period revisions.

Industrial Production (November 2009). Due for release on Tuesday, December 15th, the consensus estimate for November, per Briefing.com, is for a gain of roughly 0.5%, following a 0.1% pick-up in October. Reality is to the downside of expectations, with downside revisions to early periods a possibility. With distorted seasonals, however, almost anything is possible in the headline reporting.

Producer Price Index (November 2009). Due for release on Tuesday, December 15th, the November PPI is subject to the regular random volatility common to the series. Reporting risk generally is to the upside of expectations, though, which are around 0.8% per Briefing.com, versus 0.3% the month before.

Consumer Price Index (November 2009). Due for release on Wednesday, December 16th, consensus estimates for the seasonally-adjusted November CPI-U are around 0.4% per Briefing.com, versus 0.2% in October. Given the implied relative strength of gasoline and food prices in the November retail sales reporting, an upside surprise to consensus reporting is a fair possibility. A consensus report would boost year-to-year CPI-U inflation from minus 0.2% in October to roughly a positive 1.9% in November, a violent inflation spurt that will end the recent short-and-shallow period of formal deflation (negative year-to-year inflation).

Annual inflation would increase or decrease in November 2009 reporting, dependent on the seasonally-adjusted monthly change, versus the 1.67% adjusted monthly decline seen in November 2008.  I use the adjusted change here, since that is how consensus expectations are expressed. To approximate the annual inflation rate for November 2009, the difference in November’s headline monthly change versus the year-ago monthly change should be added to or subtracted directly from October 2009’s annual inflation rate of negative 0.18%.

Housing Starts (November 2009). Due for release on Wednesday, December 16th, November housing starts likely will show monthly variation that remains statistically insignificant, within the series’ 95% confidence interval, which was +/- 10.2% for the October report. The level of activity likely will continue bouncing along its plateau of historically low-level activity.

 __________